see below.
> but anyhow, have a look at intels SMBus specifications (if they are still
> available). it's an I²C bus for system management, on the asus tx97 there
> are the lm75, the lm78 and all SDRAM-slots connected to it. so it would
> be more logical to have an SMBus-procdevice (/proc/bus/SMBus ??) with
> submodules like lm75, lm78 or whatever you connect to it. (remember, I²C
> is an widely used industry standard.) but this way, there's the problem,
> that acops isn't realized via the SMBus/I²C of the TX (i think bx has it,
> too) but thru using some generic I/O pins of the chipset.
I guess we can't make the assumption that every monitor/watchdog is
connected through I2C, so we have to seperate the function from the bus.
> > If they use the same input/ouput via proc it is much easier to use this
> > information with mon and/or linux-ha.
> yes. how about a procdevice like /proc/sys/hardware (== file)
> containing a clear but open syntax, like "name value unit state".
> where value is !surprise! the value (numerical or ok/bad) and state is
> something like ok/bad. we need the state field, because "1236" can be
> good or bad and the reading program perhaps don't know what
> "merced-internal-temperature" means ;)
I would not use the state, but units could be something like
"boolean","on/off","1/2/4/8","2000-3000".
> > Something which simple to parse like:
> >
> > /proc/monitor/lm78:
> > Sensor-1-Temp 50 C
> > Sensor-1-Fan 4000 RPM
> > core 2.8 V
> > cpu 3.5 V
> yes, but how does joe user know if 4000 rpm means good or bad?
he is the only one who can know it and has to enter this information
into mon/linux-ha. He will have to look up this information in his
hardware manual, because it might differ depending on fan, monitor, cpu,
enclosure and probably more.
linux-ha can give him a gui for entering this information, with help of
the units this could be made nice.
damian
-- Christof Damian Technical Director http://www.mediaconsult.com/ ( btw: mediaconsult is hiring )- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu